Heredity 74 (1995)491—496 Received 30 June 1994 Genetical Society of Great Britain

Genetic diversity of wild, weedy and cultivated forms of rapa

J. H. CROUCH, B. G. LEWISt, D. J. LYDIATE AND R. MITHEN* Brass/ca and 0//seeds Research Department, John Innes Centre, Co/ney, Norwich NR4 7UJ and tSchool of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, U.K.

RestrictionFragment Length Polymorphisms (RFLPs) were used to study the genetic diversity within and between accessions of 'wild' and cultivated B. rapa. Two of the wild accessions were likely to be escapes from cultivation because of their geographical origins (Argentina and California). The nature of the other three wild accessions (from Turkey, Algeria and Sicily) was not known. Principal components analysis placed the Argentinian, Californian and Turkish accessions within a cluster which contained all the cultivated forms of B. rapa. The other two B. rapa accessions were genetically divergent and, on the basis of their RFLP genotypes, would have been considered to be more distant from the cultivated forms of B. rapa than accessions of B. nigra and B. montana. The implications of these results for germplasm conservation, selection of material for breeding programmes and phylogenetic studies on the origin of Brassica crops are discussed.

Keywords:Brassicarapa, diversity, RFLPs, wild germplasm.

taxa (Table 1). Five of the wild accessions were inter- Introduction fertile with B. rapa and on the basis of their morph- Recentadvances have been made in Brassica ology, aliphatic glucosinolate content and cytology based upon RFLPs. Comparisons between were consistent with being classified as B. rapa L. (R. F. wild and cultivated accessions of B. oleracea L. and B. Mithen and J. H. Crouch, unpublished data). A sixth rapa L. have led to hypotheses concerning the site of 'wild' B. rapa accession from Egypt (UPM 4623) was domestication of Brassica crops and their subsequent found to be B. juncea Coss. & Czern. Of the five wild spread to other geographical regions (Song et al., 1988, accessions, one was collected in Argentina and one in 1992; Warwick & Black, 1991). Wild forms of both B. California. These are very likely to be escapes from oleracea and B. rapa are difficult to distinguish from cultivation as it is highly improbable that wild B. rapa relatively recent escapes from cultivation (Mitchell, occurs in the New World. The other three accessions 1976). This has led to the inaccurate description of were collected from Turkey, Sicily and Algeria. The many accessions in germplasm collections as being accession from Turkey was collected in arable fields 'wild', whereas they are weedy remnants of previous and may also be an escape from cultivation (Oztürk et cultivations. This may lead to incorrect deductions al., 1985). The accession from Sicily was collected concerning the origin of cultivated taxa. It may also from a road verge. No information is available lead to an underestimation of the relative amounts of concerning the origin of the Algerian accession. genetic diversity within and between wild and culti- vated forms of crop species which has implications for Materialsand methods genetic conservation and the choice of genotypes for breeding programmes. Plantswere grown in John Innes No. 1 compost in To examine how incorrect identification of Brassica insect-proof glasshouses in a 18°C day/12°C night accessions may lead to problems with deducing evolu- temperature regime. Supplementary lighting was tionary relationships we obtained from germplasm provided during the winter. Total genomic DNA collections six accessions of B. rapa which were preparations were made from freeze-dried leaf described as 'wild' and compared these with cultivated material of between two and five individuals from each accession. EcoRl-restricted DNA fragments were *Correspondence separated by electrophoresis and were capillary blotted 1995 The Genetical Society of Great Britain. 491 492 J. H. CROUCH ETAL.

Table1.Originof Brassica accessions used in the present study

Accession Code no. Genome Origin Description and donor number

B. rapa subspecies rapifera 99 A Netherlands Stubble oleifera 29 A Germany Oilseed rape cv. Makelsberg oleifera 529 A Canada Oilseedrape cv. Tobin oleifera 532 A Canada Oilseedrape cv. Parkiand italica 189 A cv. Brocoletto, HRI 6218 chinensis 333 A Asia Chinese var. chinesis, IPSR# R-C-26 chinensis 699 A Asia cv. Grannaat syl vest ris 73 A Turkey Wild, UPM 6278 sylvestris 74 A Argentina Wild, UPM 5903 sylvestris 75 A Sicily Wild, UPM 6652 sylvestris 76 A Algeria Wild, UPM 6464 sylvestris 79 A California Wild, UPM 1742

Outgroups B. nigra 179 B Greece Ecotype, BGRC 34180 B. nigra 211 B Germany Wild,BGRC 32960 B. montana 196 C France Wild, BOS 12

Germplasm collections prefixes: HRI: Horticulture Research International, UK; IPSR: Institute of Science Reserch, UK; UPM: Universidad Politecnica Madrid, Spain; BGRC: Institut für Pflansenbau, Germany; BOS: Institut National de Ia Recherche Agronomique, France. Table 2. List of probes used in the present study, the Each restriction fragment was considered as a unit number of RFLPs they detected and the RFLP alleles unique character and was used to create a binary matrix. This to either B. rapa subsp. sylvestris no. 75 or no. 76 matrix was used for principal components analyses with the aid of a program provided by M. Ambrose Polymorphic Alleles specific to (John Innes Centre, Norwich) written within the Probe alleles No. 75 No. 76 GENSTAT s software (Rothamsted Experimental Station, Similarity coefficients were calculated pW2E12 8 — 1 England). — according to the Jaccard coefficient (as described in pW9A2 8 1 pW7B6 16 1 1 Gower & Legendre, 1986) for all possible pairwise pW1F6 14 — — comparisons between genotypes (Table 3). The first pW2B7 16 1 1 and second principal components were then graphi- pW2A11 16 — — cally presented (Figs 2 and 3). Similar methods of data analysis have been adopted by Wetton et al. (1987), pOlO 8 — 2 — Song etaL (1990) and Lashermes etal. (1993). p0112 10 1 p0118 7 — 1 pR54 13 — — Results 7 — I pRll5 Onehundred and twenty-three restriction fragments were scored across all accessions. The Brassica accessionstested showed a very high level of poly- morphism (Fig. 1) and no fragments were common for onto a nylon filter. Filters were probed with 11 anony- all accessions. The wild accessions from Sicily and mous gDNA probes derived from libraries of small Algeria contained several unique restriction fragments (0.8—1.6 kb) PstI-restricted fragments from B. napus (Table 2). With the exception of one unique fragment L., B. oleracea and B. rapa. The probes used for which occurred in B. rapa subsp. italica no. 189, none detecting RFLPs are listed in Table 2. Full details of of the cultivated subspecies or the accessions from DNA extraction, blotting and Southern hybridization California, Argentina and Turkey possessed any are provided by Magrath etal. (1994). unique restriction fragments.

The Genetical Society of Great Britain, Heredity, 74, 491—496. GENETIC DIVERSITY IN BRASSICA RAPA 493 of Brassica accessions used in the present study Table 3. Simiarity matrix of D values for B. nigra (2,3),B.montana (1,4,31),B.rapa subspp. rapifera (5,10), European oleifera (6,7), Canadian oleifera (8,9),italica(17,18), chinensis (16,19,20)and B.rapa subsp. sylvestris accessions collected from California Description and donor number (11,12), Argentina (13,14,15), Turkey (21,30), Algeria (22,23,24), Sicily (25,26,27,28,29)

I Oilseed rape cv. Makelsberg 32522— Oilseedrape cv. Parkiand 421 1 600003—50111 Chinese cabbage var. chinesis, IPSR 7000036— Chinese cabbage cv. Grannaat 9011801103325—0333— 10 0 0 0 0 23 3 32 — 11 01 1 0 33 3 22 2 — 12 0 1 1 0 23 3 23 2 5— 140110233331435—13 01 0 0 23 3 32 2 34— 15 0 0 0 0 33 3 12 1 3 33 3— 16 00 0 1 31 1 1 2 2 22 1 23 — 17 0 0 0 0 12 2 11 2 2 2 1 1 2 1 — 18 00 0 0 22 3 23 2 2 1 1 1 2 1 4 — 19 01 0 0 22 2 12 2 2 1 1 32 2 1 1 — — 20 0 11 0 22 2 23 1 1 1 2 22 2 1 1 5 21 0 11 1 22 2 12 1 22222 2 1 1 2 3 — Germplasm collections prefixes: HRI: Horticulture Research International, UK; IPSR: Institute of Plant Science Reserch, UK; 22 0 0 1 0 10 0 00 0 00 1 1 2 0 1 11 1 2 — UPM: Universidad Politecnica Madrid, Spain; BGRC: Institut für Pflansenbau, Germany; BOS: Institut National de Ia 23 0 1 1 1 1 0 0 00 0 1 0 1 1 1 0 1 11 1 2 8 24 0 0 0 0 10 0 00 0 00 1 0 1 11 0 0 1 2 77— Table 2. List of probes usedEach in restriction the fragmentpresent was consideredstudy, as thea unit 25 0 0 1 1 1 1 0 12 0 1 1 1 1 1 11 11 2 1 232— number of RFLPs they detectedcharacter and and the was usedRFLP to create alleles a binary matrix.unique This 26 0 0 0 1 1 1 0 11 0 1 1 1 1 1 0 0 1 0 1 2 2226— matrix was used for principal components analyses 27 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 11 0 1 1 1 1156— 1 1 1 1 0 0 1 1 0 1 1 2 with the aid of a program provided by M. Ambrose 28 0 0 1 0 01 0 11 1 22555— 29 0 0 0 0 12 2 12 0 1 1 22 1 1 0 2 1 2 2 1 105543— 30 0 0 1 0 11 2 11 0 1 2 1 1 1 1 1 1 11 2 11001001— GENSTAT s software (Rothamsted Experimental Station, 31 0 0 0 2 00 0 00 0 0 1 000 1 0 1 1 0 1 000000002— Similarity coefficients were calculated 1234567890123456789012345678901 1 2 3 cally presented (Figs 2 and 3). Similar methods of data Principal components analysisclustered the the accessions from Algeria and Sicily from each other. —analysis have been adopted by Wetton et al. (1987), accessions into several groups (Fig. 2). All seven culti- Further statistical analysis did not significantly alter the vars of B. rapa clustered together, along with the acces- results. — sions of B. rapa collected in Argentina and California, Closer examination of the principal components — suggesting, as expected, that these two accessions rep- analysis (Fig. 3) revealed grouping of taxa within the — resented weedy escapes as opposed to true wild forms main cluster containing the cultivated and weedy B. —hundred and twenty-three restriction fragments of B. rapa. The Turkish accession of B. rapa was also rapa accessions. The B. rapa subsp. oleifera (Metzg.) —were scored across all accessions. The Brassica situated close to all the cultivars. The other wild acces- Sinsk accessions were relatively distant from the B. sions from Sicily (no. 75) and Algeria (no. 76) formed rapa subsp. chinesis (L.) Makino accessions, with the —morphism (Fig. 1) and no fragments were common for two separate clusters which were more distant from the other weedy and cultivated taxa falling between these all accessions. The wild accessions from Sicily and cultivated and weedy B. rapa accessions than the out- two groups. These results are similar to those of Song mous gDNA probes derivedAlgeria containedfrom several libraries unique restriction of small fragments group represented by B. montana Pourr. and B. nigra et a!. (1990) in which Asiatic forms of B. rapa were (0.8—1.6 kb) PstI-restricted(Table fragments 2). With the exception from of one B.unique napus fragment (L.) Koch. The first principal component explained shown to be phylogentically distinct from B. rapa L., B. oleracea and B. rapa.which occurred The in B. probes rapa subsp. italica used no. 189, for none 25.5 per cent of the variation in RFLPs and separated subsp. rapifera (Metzg.) Sinsk and subsp. oleifera. The detecting RFLPs are listedof thein cultivated Table subspecies 2. Full or the details accessions fromof the wild accessions from Sicily and Algeria from the B. rapa subsp. oleifera cultivars from Canada were DNA extraction, blotting California,and Southern Argentina and hybridization Turkey possessed any other accessions. The second principal component divergent from the European cultivar (Fig. 3). Con- explained 18.3 per cent of the variation and separated siderable variation was observed within the B. rapa

The Genetical Society of Great Britain, Heredity, 74, 491—496. C The Genetical Society of Great Britain, Heredity, 74, 49 1—496. 494 J. H. CROUCH ETAL.

- 8. mo n/a,,a' t M 2 I, (era rap, t 3 11 I 4 I chinensis 14 5 I 6 7 o/e,fera 8 9 I l0 'I I i/cl/ca' 2 sylvestris, Turkey 13 I a '5 sy/vestris,California (n 16 I C 7 (n sy/vestris,Argentina 8 '9 I 20 sy/vestr/$,Algeria 21

22 11 23 24 sy/vestris,Sicily 25 26 27 I 28 chinensis 29 8. nigre Fig. 1. Southern blot hybridization of 30 EcoRl-restricted genomic DNA from 3' B. montana B .rapa, B. nigra and B. montana out- 32 groups probed with clone pWl 99.

subsp. italica cv. brocolleto (no. 189). Further analyses by the use of additional restriction enzymes) would not of diversity within cultivated forms of B. rapa require significantly alter the conclusions. Furthermore, the more extensive studies with a greater number of probes used in this study were chosen to hybridize to accessions. fragments throughout the B. rapa genome (D. Lydiate, unpublished data) in order that the results would not Discussion be biased by the detection of RFLPs within selected parts of the genome, which would be particularly Theprincipal components analysis in this study was problematical if such regions were highly conserved. based on the presence or absence of 123 restriction However, the very high levels of polymorphism (see fragments. Song et a!. (1990) concluded that the Table 2) suggest that hybridization to conserved analysis of a limited number of restriction fragments regions of the genome was not a problem. Thus, we feel (less than that used in this study) was sufficient to that the major conclusions of this study regarding the describe the major sources of genetic diversity between genetic similarities and differences between the wild Brassica accessions. It is likely, therefore, that an and cultivated accessions are valid and would not be increase in the number of restriction fragments scored significantly altered by increasing the number of (through either an increase in the number of probes or restriction fragments detected.

The Genetical Society of Great Britain, Heredity, 74,491—496. GENETIC DIVERSITY IN BRASS/CA RAPA 495

0.6 a!., 1991; Crouch et a!., 1994). Both accessions have been used in breeding programmes of B. rapa and in 0.4 the development of synthetic forms of B. napus by C 7 interspecific hybridization with B. oleracea (Crouch et 77 a!., 1994). These two accessions, therefore, seem to 11 represent forms of B. rapa which are genetically 1 1 distinct from cultivar and weedy types, and this raises 10.: 1 2 questions concerning the phylogenetic deductions -0.2 made in other studies which have used 'wild' forms of B. rapa. Sicily has been shown to be an important -0.4 6 6 6 centre of diversity for wild C genome Brassica taxa 6 6 (Mazzola & Raimondo, 1988; Snogerup et a!., 1990); -0.6 I l___ this study has indicated that it may also he an impor- -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 tant area for wild forms of B. rapa and suggests that Firstcomponent further exploration is warranted. Fig. 2. Plot of first and second principal components The other three 'wild' B. rapa accessions clustered showing relationships between all accessions tested: 1, culti- with the cultivated taxa. The accessions from Cali- vated subspecies of B. rapa; 2, B. rapa subsp. sylveslris 79 fornia and Argentina are likely to be relatively recent (California); 3, B. rapa 74 (Argentina); 4, B. rapa 73 escapes from cultivation as true wild forms of B. rapa (Turkey); 5, B. rapa 76 (Algeria); 6, B. rapa 75 (Sicily); 7, B. are restricted to the Old World. It is probable that the montana; 8, B. nigra. Turkish accession is also a weedy escape from cultiva- tion as opposed to a true wild form of B. rapa. How- ever, results such as this have led to the suggestion that wild accessions such as the Turkish accession may 7 represent the progenitors of cultivated forms of B. rapa 0.1 7 (e.g. Song Ct al., 1990). In these studies, it is likely that a comparison is only being made between cultivars and a C C 7 weedy escape from cultivation. In addition to confu- E°a 54 sion between wild and weedy accessions, it is note- 8 worthy that one of the 'wild B. rapa' accessions which 5 was examined was in fact B. juncea. 3 In studies of crosses between wild and cultivated 34 6 "-0.1 tomatoes, deVicente & Tanksley (1993) described 2 transgressive segregation in an F2 population for a number of agronomic traits. It was shown that the wild 2 species carried alleles for several agronomically -0.2 - -0.4 -0.3 -0.2 -0.1 important yield characters although this was not First component apparent from their phenotypes. It is likely that a Fig. 3. Plot of first and second principal components similar phenomenon occurs in other crop species. showing relationships between cultivated and weedy acces- RFLP analyses as described in this study may provide a sions tested: B. rapa subspecies 1, oleifera(Europe);2, means to select genetically divergent wild forms of oleifera (Canada); 3, sylvestris (California); 4, sylvestris crops, such as the B. rapa accessions from Sicily and (Argentina); 5, rap ifera; 6, italica; 7, chinensis. Algeria, in which useful alleles are present but which are not phenotypically expressed. This analysis may also provide a means to make informed choices con- The RFLPs and principal components analyses have cerning priorities for germplasm conservation and suggested that two accessions of B. rapa are genetically characterization. distant from other forms of B. rapa. This result is The confusion in identification of wild B. rapa supported by a study of secondary metabolites (Rouxel accessions is common to several other crop species. et a!., 1991) in which both of these accessions For example, wild forms of Vigna unguiculata (L.) possessed indole compounds which are not normally Walp. were once only considered to occur in West found within B. rapa. Furthermore, both accessions Africa whereas it is now known thai the greatest degree possess high levels of resistance to the fungal pathogen of genetic diversity in this species is actually found in (Desm.) Ces. et de Not. which wild forms which are abundant in southern Africa has not previously been described in B. rapa (Rouxel et (Mithen & Kibblewhite, 1993; Vaillancourt et a!., The Genetical Society of Great Britain, Heredily, 74, 491—496. 496 J. H. CROUCH ETAL.

1993). Many of the previously described 'wild' variability and relationships of Coffea species. Genet. Res. accessions of V unguiculata are naturalized cowpeas. Crop Evol., 40,91—99. The recent interest in the conservation of wild crop LETSCHERT, J. P. W. AND FESE, L. 1993. Analysis of morphological relatives has resulted in an increase in the number of variation in wild beet (Beta vulgar/s L.) from Sicily. Genet. accessions in gene banks which are described as being Res. Crop Evol., 40, 15—24. 'wild'. Many of these are likely to be recent escapes MAGRATH, R., BANO, F., MORGNER, M., PARKIN, I., SHARPE, A., LISTER, C., DEAN, C., TURNER, J., LYDIATE, D. AND MITHEN, R. 1994. from cultivation and may represent no greater levels of Genetics of aliphatic glucosinolates. 1. Side chain elonga- genetic diversity than that of cultivated forms and will tion in Brassica napus and Arabidopsis thaliana. Heredity, lead to an underestimation of the genetic divergence 72, 290—299. between crops and their wild relatives (e.g. Letschert & MAZZOLA, P. AND RMMONDO, F. M. 1988. A new species of Fese, 1993). Studies on genetic diversity using RFLPs Brassica from Sicily. Lagascalia, 15, 249—251. and other molecular markers may indicate wild MITCHELL, N. D. 1976. The status of L. subsp. accessions which are the most genetically divergent oleracea (wild cabbage) in the British Isles. Watsonia, 11, from cultivars and which contain unique alleles. These 97—103. accessions may be the most productive material for MITHEN, R. F. AND KIBBLEWHITE, H. R. 1993. Taxonomy and seeking useful agronomic characters such as pest and ecology of Vigna unguiculata in south-central Africa. disease resistance, particularly if the introgression of Kirkia, 14, 100—113. OZTURK, M., HINATA, K., TSUNODA, S. AND GOMEZ-CAMPO, C. 1985. these characters into cultivars can be aided through the A general account of the distribution of the Cruciferous use of molecular markers. in Turkey. E. U. Faculty Sci. .1. Ser. B, 6, 8 7-98. ROUXEL, T., KOLLMANN, A., BOULIDARD, L. AND MITI-IEN, R. F. 1991. Acknowledgements Abiotic elicitation of indole phytoalexins and resistance to Leptosphaeria maculans within Brassiceae. Planta, 184, Theauthors are grateful to Professor C. Gomez- 27 1—278. Campo (Universidad Politecnica Madrid) for providing SONG, K. M., OSBORN, T. C. AND WILLIAMS, P. H. 1988. Brassica seeds and for helpful discussion concerning the distri- taxonomy based on nuclear restriction fragment length bution of wild B. rapa populations, to Dr Rouxel polymorphisms (RFLPs). 2. Preliminary analysis of (INRA, Rennes) for providing seeds and to Dr M. subspecies within B. rapa (syn. campestris) and B. Ambrose for assistance with data analysis. RFLP oleracea. Theor. Appl. Genet., 76, 593—600. probes were provided by Zeneca Ltd. Technical assist- SONG, K., OSBORN, T. C. AND WILLIAMS, P. H. 1990. Brassica taxo- ance from R. Magrath, I. Parkin and A. Sharpe is grate- nomy based on nuclear restriction fragment length poly- morphisms (RFLPs). 3. Genome relationships in Brass/ca fully acknowledged. J.C. received financial support from Zeneca Ltd and the Ministry of Agriculture, and related genera and the origin of B. oleracea and B. rapa (syn. campestris). Theor. Appl. Genet., 79,497—506. Fisheries and Food. SONG, K. AND OSBORN, T. C. 1992. Polyphyletic origins of Brassica napus: new evidence based on organelle and References nuclear RFLP analyses. Genome, 35, 992—1001. SNOGERUP, S., GUSTAFSSON, M. AND VON BOTHMER, K. 1990. CROUCH, J.H.,LEWIS,B. G. AND MIThEN, R. F. 1994. The effect of A Brassica sect. Brassica (). 1. Taxonomy and genome substitution on the resistance of Brassica napus to variation. Willdenowia, 19, 271—365. infection by Leptosphaeria inaculans. P1. Breed., 112, VAILLANCOURT, R. C., WEEDAN, N. F. AND BARNARD, j.1993. 265—278. Isozyme diversity in the cowpea species complex. Crop DEVICENTE, M. C. AND TANKSLEY, S. D. 1993. QTL analysis of Sci., 33, 606—613. transgressive segregation in an interspecific tomato cross. WARWICK, S. I. AND BLACK, L. D. 1991. Molecular systematics of Genetics, 134, 58 5—596. Brassica and allied genera (subtribe Brassicinae, GOWER, J. C. AND LEGENDRE, p•1986.Metric and Euclidean Brassiceae)—chloroplast genome and cytodeme con- properties of dissimilarity coefficients. J. Classification, 3, gruence. Theor. Appi. Genet., 82, 8 1—92. 5—48. WETTON, J. H., CARTER, K. F., PARKIN, D. T. AND WALTERS, D. 1987. LASHERMES, P., CROS, J., MARMEY, P. AND CHARRIER, A. 1993. Use Demographic study of a wild house sparrow population by of random amplifed DNA markers to analyse genetic DNA fingerprinting. Nature, 327, 147—149.

The Genetical Society of Great Britain, Heredity, 74, 491—496.